As many are examining how AI-generated fortunes may reshape philanthropy, the enduring historical lesson of John D. Rockefeller for them should be more than simply his extraordinary private generosity. It should also include putting institutionalized, tax-incentivized giving in its proper place in America’s democratic order.
Staggering sums of philanthropic capital locked up in the private equity of one of the world’s most-powerful companies will soon become liquid. A technological revolution has generated wealth that, once cycled back into the charitable sector, has the potential to address many of the world’s greatest problems. We stand at the threshold of a new wave of American philanthropy. The question is whether we are ready for it.
The year was 1909, when John D. Rockefeller transferred 73,000 shares of Standard Oil Company to a private trust. The shares became liquid when Standard Oil was broken up by court order two years later; the trust became the Rockefeller Foundation, established “to promote the well-being of humanity throughout the world.” The Rockefellers themselves became the world’s most-famous philanthropic family.
An almost-identical description applies today: to the 26% stake that the OpenAI Foundation holds in its for-profit sibling; to the anticipation of a liquidity event that will unlock a new wave of philanthropy; to the entrepreneur-founders of the AI age now turning attention to their own philanthropic legacies. Some of the new AI titans view the personal generosity of Rockefeller and his peers as worthy of emulation. But the real lesson of the past runs deeper: in eras of destabilizing wealth concentration, philanthropy’s power to do good depends on institutions that are sculped through democratic scrutiny.
Brief history of a peculiar institution
The modern philanthropic foundation—think Rockefeller, Carnegie, Ford, and later Gates—is a quintessential American contradiction. We are indebted to foundations for the Green Revolution and the polio vaccine; for public libraries across the country and public-health campaigns around the world; for cultural institutions that anchor social and civic life. And yet they carry these benefits as quasi-monarchical entities that spend little, grow in perpetuity, and can entrench the priorities of a single donor’s worldview long after that donor is gone. The political theorist Rob Reich has called foundations “perhaps the most unaccountable, nontransparent, peculiar institutional form we have in a democratic society.”
The formal legal designation traces its roots to the 1969 Tax Reform Act, which imposed the basic accountability requirements still operative today: a mandatory 5% annual payout, prohibitions on self-dealing, excise taxes on investment income, and a requirement to disclose grants and financial information through annual tax filings available to the public. But the blueprint of the modern foundation was drawn 60 years earlier, when John D. Rockefeller transferred a first installment of $50 million, part of a planned $100 million gift, to a private trust administered by his son, his son-in-law, and his longtime philanthropic advisor.
It might seem surprising, given today’s sympathetic intuitions about philanthropy, that the public response to Rockefeller’s gift was not applause, but alarm. Reich recounts how Thomas Walsh, a Democratic Senator from Montana, described the proposed foundation as “a menace to the welfare of society.” A prominent Unitarian minister echoed that sentiment in testimony before Congress: however well-intentioned Rockefeller and his hand-picked trustees, the proposed entity “must be repugnant to the whole idea of a democratic society.” Both were concerned that the foundation would give the wealthiest person in the world a new and permanent instrument of public power.
What followed sets up a fascinating counterfactual in American history. Seeking a federal congressional charter, Rockefeller offered significant concessions: an asset cap of $100 million, a 50-year spend-down requirement, and public oversight of the board through trustees selected by the President, Congress, and the Supreme Court. When even these were not enough for Congress, Rockefeller turned to New York State, which granted a charter with no strings attached. American philanthropy took a fork in the road.
Whether it was the right path to take is an impossible question, requiring us to weigh indisputable public benefits against equally indisputable democratic harms. At best, we can say that the century since has vindicated some of Rockefeller’s optimism and some of Walsh’s alarm. But it is hard to deny that foundations have earned a real place in American civic life.
Recently, the history of the foundation has reached a new chapter, in which it is no longer the default vehicle of philanthropic giving. Foundations, of course, still exist: more than 100,000 of them, controlling over $1 trillion in assets. New ones continue to emerge. But among the Rockefellers of this century, the foundation is no longer the obvious choice as a philanthropic vehicle, to the extent that philanthropy is the goal at all. In its place: donor-advised funds, which offer immediate tax benefits with none of the mandatory payout requirements that foundations face; philanthropically minded LLCs, like the Chan Zuckerberg Initiative or the Emerson Collective, which permit political spending and for-profit investment alongside charitable giving; and the high-profile, individual philanthropy of donors like MacKenzie Scott, who has given away billions with minimal institutional overhead.
The obsolescence of the foundation is nothing to grieve in itself. There is nothing magical about this particular institutional formula. Norway sets a minimum capital requirement for registered foundations. India mandates 2% charitable expenditure for corporations that exceed a given size. The British charitable trust and the German Stiftungeach reflect their own institutional compromises. Hindus practice seva, Muslims zakat, Jews tzedakah, Christians charity. Philanthropy is a human constant; the institutions through which it flows are historically embedded and socially constructed, varying enormously across time and place.
Rockefeller was the world’s first billionaire. Now, private wealth touches horizons that would have been unfathomable even to him. It is reasonable to expect that old philanthropic institutions will be repurposed and new ones will emerge. But the concern that Walsh raised a century ago has not gone anywhere: will the institutions of trillionaire philanthropy be a benefit to the welfare of society, or a menace?
The Big Foundation that isn’t, and other AI hallucinations
In a widely discussed Substack essay, Nan Ransohoff argues that we are entering a third wave of American philanthropy. The first wave was the Gilded Age giving of Rockefeller, Carnegie, Ford. The second wave was the internet-era philanthropy epitomized by Gates, which repurposed the institutional form of the foundation, but with a much greater focus on measurement and evidence. The third wave is defined both by AI as the new source of wealth and by the sheer scale of money potentially coming online. “Hundreds of billions of dollars in new philanthropic capital will soon become liquid,” Ransohoff writes. The recognition, she says, “radicalized” her.
But for the spigot to actually open, the argument goes, donors need to be confident that America’s philanthropic ecosystem can absorb the new volume. What is needed, in Ransohoff’s account, are “elite capital allocators” and a new generation of philanthropic startups, supported by better incentives for top talent. This is essentially a supply-side diagnosis: latent philanthropic capital will only materialize into real giving if donors can see an adequate supply of compelling places to put it. Her back-of-the-envelope calculation suggests that roughly $50 billion annually has the potential to materialize.
The behemoth at the center of this potentiality is the OpenAI Foundation. Given OpenAI’s current valuation, the Foundation’s 26% equity stake translates to roughly $220 billion on paper. This makes it, by a considerable margin, the largest charitable entity in the world, double the size of the previous record holder—the Novo Nordisk Foundation, the philanthropic arm of the Danish pharmaceutical giant. Ransohoff speculates that the OpenAI Foundation might spend 10% of its endowment annually, once the equity becomes liquid. But even a 5% expenditure, the mandated minimum for foundations, would be a wave unto itself. The math checks out. The problem is: the OpenAI Foundation is not a foundation; it is a public charity.
The distinction is technical—but, today, has never been more consequential. It dates to the 1969 legislation, Congress’s belated response to wealthy donors parking assets in tax-advantaged endowments and disbursing little or nothing for years. The Act drew a line: organizations with broad public support (e.g., the Red Cross and the ACLU) became public charities, lightly regulated on the theory that a broad funding base is itself a form of accountability. Organizations funded by a small number of donors and built primarily to give money away (e.g., Rockefeller, Carnegie, Ford) became private foundations, bound by mandatory payout requirements and greater transparency obligations. Colloquially: foundations give money away, whereas public charities do things with money given to them. Both can hold vast endowments. Only foundations are required to spend from theirs.
OpenAI was incorporated as a public charity in 2015 because that is what it was: a research organization with a charitable purpose, funded by a diverse group of donors, designed to do something in the world rather than write checks to others who would. Its original mission, reconstructed from early filings by Simon Willison, committed to advancing AI “unconstrained by a need to generate financial return,” building safe AI technology, and sharing “plans and capabilities along the way.” Nine donors contributed a combined $1 billion at founding. The organization noted in its 2016 press release that it expected to “only spend a tiny fraction of this in the next few years.”
What happened next is well-known in broad strokes: the gradual transformation of a nonprofit research lab into one of the most-valuable private companies in human history. The mission statement has been revised six times in nine years. The word “safely” has been quietly removed. A controversial restructuring in October 2025 transferred three-quarters of the nonprofit’s equity to private investors, with the public charity—now renamed the OpenAI Foundation—retaining a 26% minority equity stake in the for-profit entity it nominally oversees.
The renaming reflects an institutional sleight-of-hand. By calling itself a foundation, OpenAI borrows the legitimacy that Rockefeller, Ford, and Gates spent a century building. This legitimacy was earned, however imperfectly, through democratic debate. We assume, because it uses the word, that the OAI Foundation will behave as foundations behave, including through mandatory disbursements and a minimal degree of public accountability. But those assumptions are not legally warranted. The OAI Foundation is an AI hallucination: a confident, fluent, entirely plausible-sounding entity that does not correspond to anything real.
That hallucination appears even before the 2025 restructuring and renaming. Consider OpenAI’s 2024 tax returns, the most recent publicly available. The nonprofit’s cumulative revenue over the prior five years was approximately $9 million, with $7.5 million coming from public sources—far more than the 33% threshold required to maintain public-charity status. But that is only because the real money, including billions in revenue generated by ChatGPT, was siphoned off to for-profit subsidiaries. The nonprofit had already been marginalized within the larger corporate structure. The disingenuous picture that emerges is of a relatively modest organization—larger than a local historical society, but smaller than a community college. The 2024 returns document a single employee who worked 10 hours a week, earning $13,000 over the full year.
Sen. Walsh, were he alive today, would be howling. Where are the congressional hearings? How is it that an entity reporting a single employee in 2024 is, one year later, the largest charitable organization in the history of the world? How is this not a menace to society? A legal challenge mounted in 2024 did, in fact, accuse OpenAI’s executives of having “stolen a charity,” betraying the public’s trust by converting a nonprofit research lab into a behemoth private enterprise. Walsh might have been encouraged by the lawsuit, only to fall further into despair upon learning of the plaintiff: Elon Musk, himself building a rival AI company, soon to become the world’s first trillionaire. The lawsuit was dismissed in under two hours—not on the merits of the suit, but because Musk had waited too long to sue.
It is worth noting that a liquidity event does not materially change the picture. The OpenAI Foundation could spend vast sums today if it chose to do so, taking out loans against its equity or through countless other mechanisms. It has already begun to make some modest philanthropic expenditures, such as grants totaling $50 million in 2025 through its People-First AI Fund and a commitment of at least $1 billion in the year to come. But even that billion, if it materializes, would represent less than half of one percent of the Foundation’s paper wealth. OpenAI going public would make it easier for the Foundation to access that wealth, but it would create no obligation to do so.
How does a clearer understanding of the OpenAI Foundation change our view of the third wave of philanthropy that Ransohoff describes? First, it complicates her back-of-the-envelope calculation, as others have noted. Ransohoff’s estimate of $50 billion in new philanthropic capital rests heavily on the assumption of 10% expenditure—$22 billion—from the OAI Foundation alone. The actual track record of OpenAI’s giving suggests that this number might vastly overestimate the foundation’s future giving. On the other hand, Ransohoff may also be underestimating the philanthropic flows by an order of magnitude. Perhaps the foundation decides that, given the stakes of the next few years, 20% expenditure is more appropriate. Perhaps OpenAI quadruples in value.
The unknowability of the OAI Foundation’s future giving is itself the most-important data point, and it suggests that we are asking the wrong question. Whether the nonprofit sector is ready for a new wave of philanthropic capital is important. But whether American democracy is ready matters far more.
The real lesson
Should we welcome the third wave of philanthropy that Ransohoff sketches out? “The opportunity for good created by this new wave of wealth is enormous,” Ransohoff writes. She is right. The challenges are real and urgent: climate change and pandemic preparedness; alleviating suffering through access to medication that has long existed, and alleviating it further through new frontiers of biomedical research; of course, AI safety itself. Philanthropy has addressed many of these challenges in the past, and it has the potential to address all of them in the future. There is also something genuinely appealing about the vision of a philanthropic ecosystem that draws top talent towards great societal challenges, rather than toward AI shoe companies. Elsewhere, Ransohoff writes of a new class of “general managers” of the world’s most-formidable problems. Yes, please.
One of the world’s most-formidable challenges today, including and especially in the United States, is a crisis of democracy. The crisis appears as a collapse of trust in our representative institutions, which become less deserving of trust as they fail to represent us; as the blatant corruption of a hollowed-out, captured government; as failure thus far “to find our way to accommodations with one another so that we can birth for ourselves a sense of shared fate.” Among the many causes of this crisis: the widely acknowledged destabilizing effect of extreme wealth concentration that has reached historic levels and that AI is now accelerating.
Dario Amodei is one of the few prominent voices in the AI world to name this threat directly. In his January 2026 essay “The Adolescence of Technology,” Amodei identifies the concentration of economic power as one of the central risks of the AI age, equally deserving of alarm as the more-commonly discussed dystopias: of machine overlords that clash with its creators, of the rogue actor that acquires capacity for destruction previously reserved for nuclear-armed states. In addition to these “security risks,” Amodei insists on highlighting the quieter, more-structural risk that AI simply makes the rich incomprehensibly richer, concentrating wealth and power in ways that lead democracies inevitably toward heart attack. It is worth quoting at length:
But another kind of disempowerment can occur if there is such a huge concentration of wealth that a small group of people effectively controls government policy with their influence, and ordinary citizens have no influence because they lack economic leverage. Democracy is ultimately backstopped by the idea that the population as a whole is necessary for the operation of the economy. If that economic leverage goes away, then the implicit social contract of democracy may stop working. … I worry that it is already starting to happen.
Macroeconomic interventions such as more-aggressive taxation are part of the suite of solutions to the crisis of wealth and power concentration. But it is notable that, on Amodei’s account, philanthropy also has a meaningful role to play. Descrying the all-too-common attitude in Silicon Valley that “philanthropy is inevitably fraudulent or useless,” Amodei instead points approvingly toward the Gilded Age philanthropy of Rockefeller and Carnegie, “who felt a strong obligation to society at large, a feeling that society had contributed enormously to their success and they needed to give back.” He continues: “That spirit seems to be increasingly missing today, and I think it is a large part of the way out of this economic dilemma. Those who are at the forefront of AI’s economic boom should be willing to give away both their wealth and their power.”
For their part, Amodei and his Anthropic co-founders have committed to donating 80% of their wealth, and Anthropic runs an exceptionally generous matching-donation program for its employees. These commitments, too, inform Ransohoff’s back-of-the-envelopment calculations for the third wave of American philanthropy. In contrast to the OAI Foundation’s murky billions, they feel credible. The intentions seem genuine. But that doesn’t make the institutional logic any less flawed. Amodei is right to admire the “strong obligation to society at large” felt by Rockefeller and his peers, and right to lament its absence from much of the current tech landscape. But the real lesson of Rockefeller philanthropy is not about the donor’s intentions. It is that philanthropy is as likely to concentrate power as to disburse it, and the difference depends on the institutional structures through which the giving flows.
Congress in the early 20th Century recognized this two-sidedness of philanthropy: its capacity to fund valuable projects in society alongside its tendency to concentrate the power of those deciding which projects and purposes are worth funding. A wealthy individual who buys expensive artwork or a luxury yacht is exercising less power over public life than one who donates a billion dollars to, say, education or health. Ross Douthat exhorts the new AI donors to spend money on beauty: “Build monuments, statues, museums, universities, cathedrals, public gardens—and yes, even mansions for yourself.” One reason to find this appealing, beyond the obvious, is that getting beauty wrong is less harmful than misguided expenditures on causes seen as more philanthropically substantive. Better an ugly museum than an education campaign that creates a mess for others to fix.
The foundation as conceived in the early 20th Century and amended in 1969 is an institutional settlement designed to address exactly this difficult trade-off: how do we encourage philanthropic spending that has the potential to benefit society without overly exacerbating the concentration of power that, in itself, is a menace? No one would claim the particular settlement represented by the foundation is perfect. Perhaps it is not even very good. But as it becomes increasingly obsolete, we should not forget the original problem it was designed to solve. What the new wave of philanthropy needs, more than new nonprofit startups and elite capital allocators, is an entirely new imagination of the institutions that mediate how money flows from its wealthiest members back toward society.
The past can help get us started. The guardrails that Congress envisioned over a century ago are worth dusting off and considering anew: caps on endowment size, mandatory spend-down requirements, public representation on foundation boards. But the future will also require a radically new institutional imagination. For example, we should be willing to draw entirely new regulatory boundaries: not just between public charities and private foundations, as the 1969 Act does, but between small and large philanthropic entities. Community foundations and modest family foundations need not be regulated in the same way as an entity holding $220 billion in assets. The growing prominence of for-profit entities as vehicles of giving likewise invites us to more fundamentally reassess what philanthropy means in a legal sense. As an easy initial effort, we can reform or eliminate donor-advised funds, which have taken on a role in the philanthropic landscape vastly different from the original intentions behind their design.
Life, liberty, and the pursuit of civilizational flourishing
The challenge of democratically redesigning the institutions of philanthropy for the AI age is compounded by a final consideration. Though Amodei may be an exception, many of the new wealth-holders of the AI age appear to view democracy not as a value to be preserved, but as an obstacle to be removed. An actual antipathy toward democracy is a defining feature of accelerationism, the loose but influential ideology circulating in Silicon Valley. More broadly, democracy seems to inspire a kind of benign indifference among the new donor class: a general sense that it will sort itself out once the real existential risks have addressed.
That is a dangerous mistake. Ransohoff almost sees this when she directs attention toward a horizon that reaches beyond today’s most-urgent challenges. Distinctive of the third wave of American philanthropy, she argues, are questions of “civilizational flourishing, meaning, and what makes a life good.” In some sense, this is the broader promise of AI itself: the horizon that makes the hard work of navigating the adolescence of technology worth undertaking. It is what we have most reason to be excited about.
But flourishing requires political agency, the capacity of ordinary people to shape the conditions of their own lives and to participate meaningfully in the decisions that govern them. And if we wait until the carbon has been sucked from the air and the AI demons exorcised from the machines before figuring out how to live with each other democratically, it will be too late. The institutions that make democratic life possible cannot be renovated after the fact. They have to be treated with the same urgency as the other problems that AI philanthropy views as existential.
Two hundred and fifty years ago, a group of men who signed their names with improbable flourishes spoke of pursuit of happiness, rather than civilizational flourishing; liberty, in place of meaning; life, too. Call it life, liberty, and the pursuit of happiness. They tried to design institutional structures that would secure these things, especially cognizant of how power needs to be distributed and balanced, lest it become overly concentrated. We still haven’t figured it out. But the intuition that decisions should be made by a demos, rather than a general manager, seems worth preserving.
Does Silicon Valley have a word for that?
***
Jeremy McKey is a tech policy researcher at Princeton University’s Center for Information Technology Policy and a 2025-26 policy fellow at the Harvard Kennedy School’s Allen Lab for Democracy Renovation. Previously, he was director of special projects at the Rockefeller Brothers Fund.